THE JOURNAL OF COMPARATIVE NEUROLOGY 322:501-518 (1992)
Selective Vulnerability of the Hippocampal Pyramidal Neurons to Hypothyroidism in Male and Female Rats M.D. MADEIRA, N. SOUSA, M.T. LIMA-ANDRADE, F. CALHEIROS, A. CADETE-LEITE, AND M.M. PAULA-BARBOSA Department of Anatomy, Porto Medical School (M.D.M., N.S., M.T.L.-A., A.C.-L., M.M.P.-B.) and Division of Mathematics and Physics, Department of Civil Engineering, Faculty of Engineering (F.C.), Porto, Portugal
ABSTRACT Thyroid hormone deficiency has long been considered to affect profoundly such cognitive functions as learning and memory, which are known to depend on the structural integrity of the hippocampal formation. Since we previously found that the number of granule cells of the dentate gyms is reduced in hypothyroid animals, we decided to extend our observations to the pyramidal cells of the hippocampus in order to gain further insight into the effects of hypothyroidism upon the other neuronal links of the hippocampal trisynaptic circuitry, inasmuch as CA1 neurons are known to be particularly vulnerable to aggressive agents. Groups of 6 male and 6 female rats aged 30 and 180 days were analysed separately after being treated as follows: (1) hypothyroid from day 0 until day 30 (30-day-old hypothyroid group); (2) respective 30-day-old control; (3) hypothyroid from day 0 until day 180 (180-day-old hypothyroid group); (4)hypothyroid until day 30 and thenceforth maintained euthyroid (recovery group); (5) hypothyroid since day 30 (adult hypothyroid group); and (6) respective 180-day-old control. The volume of the pyramidal cell layer of the CA1 and CA3 regions and the numerical density of the respective neurons were evaluated, thereby allowing us to estimate the total number of pyramidal cells in each hippocampal region. The areal density and the mean nuclear volume of CA1 and CA3 pyramidal cells were also estimated. In the CA3 region, we found that hypothyroidism, whatever its duration and time of onset, induces a reduction in the volume of the pyramidal cell layer and a parallel increase in the numerical density of its neurons, without interfering with the total number of pyramidal cells. Conversely, in the CA1 region, thyroid hormone deficiency started either neonatally or during maturity was found to lead to a decrease in the total number of pyramidal cells. Reductions ranging between 14.2 and 22.5% were found in 30 and 180-day-old hypothyroid groups. The reestablishment of a euthyroid state did not ameliorate the referred neuronal loss. The present results support the view that hypothyroidism induces small alterations in the structural organization of the hippocampal CA3 region, contrary to what happens in CA1 in which neuronal death occurs. Furthermore, the data presented herein demonstrate that the total number of CA1 pyramidal cells displays sexual dimorphism that is not affected by thyroid hormone manipulations. These findings allow us to admit that hypothyroidism induces anatomical changes in the hippocampus which are in line with those reported for the dentate gyrus, thereby providing additional morphological basis for the cognitive alterations described in this condition. o 1992 Wileg-Liss, Inc. Key words: thyroid deficiency, hippocampus, pyramidal cells, neuronal death, sexual dimorphism
In humans, the exposure to low levels of thyroid hormones during late fetal or early postnatal periods commonly leads to developmental abnormalities and mental deficiency (Burgi, '86; Fisher, '89). The relationship between cognitive deficits and thyroid hormone deficiency became apparent after the report Of Kerley ('36) about childhood myxedema. Subsequently, additional work reinO
1992 WILEY-LISS. INC.
forced this view by further demonstrating that the cognitive and sensory alterations that are hallmarks of congenital hypothyroidism are either irreversible or only partially Accepted ~ ~29, 1992, ~ i l Address reprint requests to M.D. Madeira, Department ofAnatomy, Porto Medical School, Alam. Prof. Herngni Monteiro, 4200 Porto, Portugal.
M.D. MADEIRA E T AL. reversible, even when appropriate replacement therapy is promptly started (Hulse, '84; Biirgi, '86; Fisher, '89). The search for clues to the nature of this dramatic situation justifies the intensive research that, during the past decades, has been conducted in several regions of the central nervous system (CNS) under hypothyroid conditions. From the wealth of work produced, it became evident that the presence of normal levels of these hormones is indispensable for the harmonious development and subsequent maintenance of the structural organization of the CNS. Most of the morphological work produced in this domain was performed in the cerebellum (Nicholson and Altman, '72a,b; Legrand, '82-'83; Madeira et al., '88a) and in the neocortex (Eayrs and Taylor, '51; Eayrs, '55; RuizMarcos et al., '79, '83). In contrast, studies designed to evaluate the effects of thyroid hormone deficiency upon the structure of the hippocampal formation are scarce, which is hard to understand, since the hippocampal formation plays a role of paramount importance in a wide range of cognitive functions that are known to be deeply affected in hypothyroidism, namely learning processes and memory tasks (Strehler, '89; Alkon et al., '911, especially those including spatial information (McNaughton et al., '89; Alkon et al., '91). Experiments that used simple learning tests (Eayrs, '61; Eayrs and Levine, '63; Davenport and Dorcey, '72) succeeded in demonstratingthat in animal models, hypothyroidism exerts deleterious effects upon learning capacities; i.e., it induces behavioral alterations similar to those observed in humans. This is likely because most of the morphological alterations observed in the CNS of hypothyroid animals parallel those found in human studies performed in patients with thyroid hormone deficiency (Gilroy and Meyer, '75). The hippocampal formation is particularly suitable for morphometric studies because it is a relatively simple cortical structure (West et al., '78; Stanfield and Cowan, '79; Amaral and Witter, '89). Still, the existing quantitative studies regarding the effects of hypothyroidism upon its morphological organization are almost exclusively restricted to the granule cells of the dentate gyrus (Rami et al., '86b; Madeira et al., '88b, '91b), which are known to have a peculiar neurogenesis (Bayer and Altman, '74; Bayer, '80; Altman and Bayer, '90a). We have unequivocally demonstrated that hypothyroidism, starting either in the neonatal period or in adulthood, leads to a reduction in the total number of these neurons, which is not restored by the normalization of thyroid hormone levels (Madeira et al., '88b, '91b). We were able to conclude that the mechanisms that underlie those changes were multiple: in addition to the widely accepted interference with granule cell migration (Lauder, '79; Legrand, '82-33; Rami et al., '86b), thyroid hormone deficiency also alters granule cell proliferation and induces iieuronal degeneration (Nicholson and Altman, '72a; Legrand, '82-'83; Madeira et al., '91b). Bearing in mind that granule cells are just the first link of the hippocampal circuitry (Amaral and Witter, '89), we thought it would be worthwhile to extend our observations, using the same experimental paradigm, to the remaining components of the aforementioned circuitry. Thus, we centered our studies on pyramidal cells of hippocampal regions CA1 and CA3 that display a neurogenic pattern that is markedly different from that of granule cells as it occurs predominantly during the prenatal period (Bayer and Altman, '74; Bayer, '80; Altman and Bayer, '90b). Neurogenesis is o f paramount importance for the understanding of
hypothyroid-induced alterations because in neurons generated prenatally, the alterations mainly derive from changes in the process of differentiation and thus are confined to cell neurites, becoming apparent on the components of the neuropil (Ruiz-Marcos et al., '79, '83; Madeira et al., '90, ' 9 1 ~ ) Conversely, . in neuronal populations that undergo a significant postnatal formation, besides those alterations (Legrand, '82-'83; Rami et al., '86a), a reduction in the number of neurons is likely t o occur, since both neuronal proliferation and migration are also affected by the hormone deficiency and thereby interfere with the process of cell acquisition (Rami et al., '86b; Madeira et al., '88a,b, '91b). The estimators chosen to be evaluated were the total number of CA1 and CA3 pyramidal cells, their respective numerical density, and the volume of the pyramidal cell layer, both in young (30-day-old) and adult (180-day-old) animals submitted to different experimental conditions of thyroid hormone deficiency. These determinations were undertaken separately for CA1 and CA3 regions since their respective pyramidal cells exhibit a selective vulnerability to various aggressions (Nadler et al., '78; Griffiths et al., '84; Bertram et al., '90; Represa et al., '91; SchmidtKastner and Freund, '91 ). Moreover, as hippocampal neurons display receptors for gonadal hormones (Stumpf and Sar, '78; Loy et al., '88) and hypothyroidism alters the entire hormonal milieu (Dumm et al., '851, we attempted to analyse whether the total number of pyramidal neurons displays sexual dimorphism either under normal conditions or following thyroid hormone manipulations. With this quantitative investigation, an extension of the study undertaken in the dentate gyrus under hypothyroid conditions, we expect to get a better insight into the effects of hypothyroidism in the hippocampal formation by answering the following questions: (1)does hypothyroidism that starts during the neonatal or adult periods interfere with the number of pyramidal cells?, (2) does long-term hypothyroidism induce progressive cell death?, (3) is there any morphological improvement once euthyroidism is reestablished?, (4)is there any regional vulnerability to hypothyroidism among CA1 and CA3 pyramidal cells?, and (5) are there sex differences among CA1 and CA3 pyramidal cells in normal and hypothyroid conditions?
MATERIALS AND METHODS Animals and treatments Some of the animals herein employed were used in a previous work (Madeira et al.,'91b). Data were collected from 30 and 180-day-old SpragueDawley rats from the colony of the Gulbenkian Institute of Science (Oeiras, Portugal). Males and females were separately analysed. On the day of birth (day PO) the litters were equalled to 8 (4male and 4 female). Separate groups of 6 male and 6 female rats, formed by pooling animals from at least 5 different litters, were treated as follows: Thirty-day-old hypothyroid groups. Hypothyroidism was induced from day PO until day P30 by a subcutaneous daily injection of propylthiouracil (PTU): 0.05 mlO.2% PTU on days PO-P10, 0.1 mlO.2% PTU on days Pll-P20, and 0.1 mlO.4% PTU on days P21-P30 (Nicholson and Altman, '72a; Madeira et al., '91b). One-hundred eighty-day-old hypothyroid groups. Rats were treated as in the previous group from day PO untjl day P30. At day P30, a thyroidectomy was surgically performed
HYPOTHYROIDISM AND HIPPOCAMPAL PYRAMIDAL NEURONS with the aid of a dissecting microscope after intraperitoneal anaesthesia (2 mlikg body weight) with a solution of 2.5% sodium pentobarbital in physiological saline (Madeira et al., ’91b). Recovery groups. Hypothyroidism was restricted to the first postnatal month. The newborn rats were rendered hypothyroid from day PO until day P30 by administering PTU as previously described. Thence, they were allowed to survive without further treatment. Adult hypothyroid groups. Rats were maintained in standard conditions as the controls until day P30 and then hypothyroidism was induced by surgical thyroidectomy. Thirty and 180-day-old control groups. Rats were not submitted to any kind of treatment during both experimental periods. For the sake of consistency, each experimental group should have its respective control group subcutaneously injected with physiological saline and/or sham-operated, as animal manipulations might increase the blood levels of stress hormones known to exert deleterious effects upon neurons (Sapolsky, ’87). However, the glucocorticoidinduced neuronal loss in the hippocampal formation only occurs following cumulative hormonal exposure over the lifespan (Sapolsky, ’87). Thus, we decided not to include those controls because enlarging the number of groups to be studied would render very difficult to get an overall view of the effects of hypothyroidism. In addition, our own experiments (Madeira et al., ’91b), as well as data from other authors (Matsumoto et al., ’911,clearly show that this kind of animal manipulation does not alter the number of the hippocampal neurons. As males and females were separately studied, 12 groups of 6 animals each were analysed. The animals were housed at a light-dark cycle of 12/12 hours and after weaning they had free access to standard laboratory diet and water. Body weights were recorded every third day during the first month of experiment, and thereafter every fifteen days; the last determination was performed on the day of sacrifice.
Hormonal determinations Serum thyroxine (T4) levels were determined by radioimmunoassay by using 50 pl blood samples and commercial kits supplied by Pharmacia. Though the assays of triiodothyronine, reverse triiodothyronine, and thyroid-stimulating hormone would be desirable for a full characterization of hypothyroidism, they were not performed because specific immunoreactants were not commercially available. Yet, T4 concentration is an accurate estimator of hypothyroidism and thus the obtention of a complete profile of the thyroidal hormone status was not considered to be an absolute requisite.
General procedures At the end of each experimental period (30 or 180 days), animals were anaesthetized with ethyl ether and, after heparinization, blood samples (1 ml) were collected from the heart for hormonal assays. The rats were then transcardially perfused with a solution of 1%glutaraldehyde and 1% paraformaldehyde in 0.12 M phosphate buffer at pH 7.2. The thyroidectomized rats (180-day-old hypothyroid and adult hypothyroid groups) were carefully searched for the presence of thyroid gland remnants and when any evidence of thyroid tissue was present, the animals were discarded.
503
In the remaining groups, the thyroid glands and the adjacent portion of the trachea were isolated and weighed. The brains were removed from the skull and weighed. The cerebral hemispheres, after being separated by a longitudinal section in the midsagittal plane, were immersed for 2 hours in fresh fixative solution. The left hippocampal formations were isolated by dissection from the adjacent structures and weighed.
Volumetric estimations After removal of the frontal and occipital poles of the right hemispheres, the blocks of tissue containing the hippocampal formations were dehydrated in an ascending series of ethanoliethanol-ether and embedded in celloidin (Fig. 1A). After the embedding procedure, the blocks were serially sectioned in the horizontal plane at a nominal thickness of 60 pm with a sliding microtome. The sections from each hippocampal formation were mounted in glass slides and stained with cresyl violet (Fig. 1B). The actual section thickness, evaluated by differential focusing with a x 100 oil immersion objective lens, was found to be 59.3 ? 6.4 pm. The volume of the pyramidal cell layer of the CA1 and CA3 regions of the hippocampus was estimated from sections covering the entire dorsoventral extent of the hippocampal formation (Fig. 1B). Because the morphology of the hippocampal components changes markedly from section to section at both septal and temporal poles, the first 25 septal and the last 10 temporal sections were studied. At the midseptotemporal level, where the profile of the hippocampal components changes very little, sections studied were 240 pm apart; i.e., only every fifth section was selected for volumetric determinations. The boundaries of the pyramidal cell layer were delineated according to the criteria described in detail by West et al. (’78). Although the terminology used to designate the hippocampal regions followed in this paper is the one described by Lorente de N6 (’341, in the pyramidal cell layer only two regions were considered: CA1 and CA3 (Fig. 1B).The distal border of the first one was considered at the point where the densely packed pyramidal cell layer gives rise to the wider cell layer of the subiculum (Figs. lB, 2A,B). The discrimination between CA1 and CA3 was made taking as reference the CA2 region (Fig. 2A,C) which is easy to define by the abrupt transition in cell size and cell layer organization (Lorente de NO, ’34).For volumetric estimation purposes, this small transitional zone was included in CA3 because the transition between CA2 and CA3 is hardly detectable in conventionally stained sections. In addition, the neurons located in the pyramidal cell layer within the hilus of the dentate gyrus, which can be clearly differentiated from the hilar cells (Lorente de N6, ’34; Amaral, ’781,were also considered as belonging to the CA3 region. From all the selected sections, the boundaries of the pyramidal cell layer of the CA1 and CA3 regions were drawn using a camera lucida attachment at a magnification of x63 and their areas were measured with a MopVideoplan. Based on these measurements, the volume of the pyramidal cell layer of both hippocampal regions was calculated according to the Cavalieri’s principle using the “arithmetic mean” approximation given by the equation: n-l
V = S F , i=x1
A,
+ Ai+l
di,
M.D. MADEIRA ET AL.
504 in which SF, is the tissue shrinkage factor, A, is the surface area of the respective pyramidal cell layer in the iJh section, n is the number of sections measured, and d, is the distance between areas A, and A,+1(Uylings et al., '86). The SF, was estimated separately for male and female rats at the septal and temporal poles of the dissected left hippocampal formations. After removing the midseptotemporal part of the hippocampal formations, a section was immediately obtained in the horizontal plane and by using an Oxford vibratome from the surface of each pole facing the site of transection; the cross-sectional area of these sections was then measured. The remaining portions of both poles were embedded in celloidin, and the area of the first section obtained from both blocks, immediately adjacent to the vibratome section, was also measured. The referred areal measurements were made with the aid of a Mop-Videoplan after photographing the respective sections on a stereoscopic microscope. The SF, was then calculated from the areas of the two immediately adjacent sections according to the formula:
in which c is a shape constant, A, is the vibratome section area, A, is the celloidin section area, and V, and V, are the volumes of the sections obtained with the vibratome and after embedding in celloidin, respectively (Uylings et al., '86).
Quantifications of neuronal parameters
sections selected at random from different blocks and cutting 3 cross-sections of each at the same thickness. Three measurements were made in each transverse section and the mean thickness determined was 2.02 2 0.20 pm. All neuronal estimations were performed in the subregion c of CA3 (Lorente de NO, '34) and in the proximal part of CA1 (Fig. 2A). However, because the methodology applied was identical for control and experimental groups, there is no reason to assume that this might have interfered with the end results of a comparative study such as this. Mean nuclear uolume. The size of the pyramidal cell nuclei was estimated using the nucleator (Gundersen, '88; Gundersen et al., '88; Madeira et al., '91b). Photographs of the same area of the pyramidal cell layer of the CA1 and CA3 regions were obtained from sets of semithin sections of the hippocampal formation, and then analysed at a final magnification of ~ 1 , 8 6 0 (Fig. 2D). The nuclei to be measured were selected by applying the disector (Sterio, '84; Madeira et al., '88a,b, '91b). The estimations were carried out in three-dimensionally isotropic directions using a systematic set of points (Cruz-Orive and Hunziker, '86; Gundersen et al., '88) and a bidirectional ruler equidistant in length cubed (Gundersen and Jensen, '85). Taking into account the ruler construction and the magnification of the photographs (Braendgaard and Gundersen, '861, the mean volume of the nuclei),W was estimated according to the equation:
4T -
VN -- -3e ; ,
For neuronal quantifications, the left hippocampal formations were used. They were sectioned in the horizontal plane using a tissue chopper and the blocks processed, at room temperature, as though being prepared for electron microscopic observation (Palay and Chan-Palay, '74). After postfixation in a solution of 2% osmium tetroxide in 0.12 M phosphate buffer, the blocks were rinsed in 25% ethanol and then dehydrated through a graded series of ethanols. Block-staining took place at the 70% stage with 1%uranyl acetate during 1 hour. After passage through propylene oxide the material was embedded in Epon. Because the neuronal density and size vary throughout the septotemporal axis of the hippocampal formation (Gaarskjaer, '781, and because there is evidence that under some experimental conditions the vulnerability of hippocampal neurons changes along the septotemporal axis (Ashton et al., '89), a systematic random sampling procedure (Gundersen and Jensen, '87) was applied to select the sections used for the quantification of neuronal parameters. Therefore, 6 blocks were selected from each left hippocampal formation, and froin each block, 2 groups of 3 consecutive 2 pm semithin sections were obtained, mounted in glass slides, and stained with toluidine blue. The actual section thickness was determined after reembedding 10 semithin
in which is the mean of all cubed distances measured (Gundersen, '88). Areal density. The number of cells per unit surface area of the pyramidal cell layer-areal density (NA)-was calculated by counting, at a final magnification of ~ 3 2 0 the , number of existing neurons within the boundaries of a previously determined area of the pyramidal cell layer in the CA1 and CA3 regions. For this purpose, the first and the last sections of each group of 3 consecutive semithin sections were used per animal. Numerical density. The number of cells per unit volume of the pyramidal cell layer-numerical density (Nv)was estimated by applying the disector (Sterio, '84; Madeira et al., '88a,b, '91b). From all groups of semithin sections obtained, photographs of the same area of the pyramidal cell layer of the CA1 and CA3 regions were taken from two 0 consecutive sections at a final magnification of ~ 3 2 (Fig. 2D). A total of 24 disectors was made per animal. Total number of cells. The total number of cells of the hippocampal CA1 and CA3 regions was calculated by multiplying the volume of the pyramidal cell layer of each region, after correction for SF,, by the numerical density of the respective neurons. For technical reasons, it was not possible to estimate these parameters in the same hippocampal formation, which is known to be of great impor-
Fig. 1. A: Lateral view of the right hippocampal formation from a 180-day-old male control rat after removal of the neocortex. The lines (a,b,c,d)indicate the location from where the sections present in B were obtained. x 5 . 3 . B: Light micrographs of horizontal celloidin sections taken from representative levels of the right hippocampal formation whose location is indicated by the lines traced in A. The borders
between CA1 and CA2-3 regions are indicated by arrows, whereas the borders between CA1 and the subiculum are indicated by arrowheads The CA1 and CA3 regions of the pyramidal cell layer from those sections can be seen, at a lower magnification, in the correspondent schematic camera lucida drawings. CA1 and CA3, regions of the hippocampus; DG, dentate gyrus; S, subiculum. x 18.4.
Figure 1
A
Figure 2
HYPOTHYROIDISM AND HIPPOCAMPAL PYRAMIDAL NEURONS tance to get unbiased results. Still, for comparative purposes, the results obtained can be considered reliable because the methodology employed was the same for all groups studied. With the histological processing used, the discrimination between pyramidal and nonpyramidal cells is impossible to be achieved and, therefore, the results obtained are estimates of the total number of neurons in the pyramidal cell layer of each hippocampal region. Since pyramidal cells are the predominant cell type in the pyramidal cell layer and the nonpyramidal neurons account just for a very small proportion of the cells in this layer (Lorente de N6, '34; Schlander and Frotscher, '86; Danos et al., '91), it is reasonable to assume that the results obtained are representative mainly of the former cell type.
Statistical analysis Normal quantile plots were used as a tool for assessing the normal distribution and the variance homogeneity of the data (Moore and McCabe, '89). Due to the normality and variance homogeneity of the sampled populations, a single two-way analysis of variance (ANOVA) was carried out on data from control and experimental groups to discern main effects. Treatment and sex were used as independent variables and the animals as replicates. The remainder mean square was used as the error term. Post hoc linear polynomial contrasts (Moore and McCabe, '89) were applied to test whether group means differed significantly from each other; painvise comparisons were performed. A Multistage-Bonferroni test (Moore and McCabe, '89) was used to control for unacceptable levels of type I error, thus obtaining a stronger protection for false rejections of the null hypothesis. Results for which the null hypothesis would be rejected according to the uncorrected procedure applied are regarded as borderline. Throughout this study, values are expressed as mean 5 S.D.
RESULTS Body weights
507
TABLE 1. Results of Two-way ANOVA on the Body, Brain, Hippocampal Formation, and Thyroid Gland Weights of the 30-Day-OldAnimals Treatment (df 1,20)' F value Body weight Brain weicht Hippocamid formationweight Thyroid gland weight
Sex (df 1,201
P value
Interaction (df 1,201
F value P value F value P value
132.72 5 1 39
Selective Vulnerability of the Hippocampal Pyramidal Neurons to Hypothyroidism in Male and Female Rats M.D. MADEIRA, N. SOUSA, M.T. LIMA-ANDRADE, F. CALHEIROS, A. CADETE-LEITE, AND M.M. PAULA-BARBOSA Department of Anatomy, Porto Medical School (M.D.M., N.S., M.T.L.-A., A.C.-L., M.M.P.-B.) and Division of Mathematics and Physics, Department of Civil Engineering, Faculty of Engineering (F.C.), Porto, Portugal
ABSTRACT Thyroid hormone deficiency has long been considered to affect profoundly such cognitive functions as learning and memory, which are known to depend on the structural integrity of the hippocampal formation. Since we previously found that the number of granule cells of the dentate gyms is reduced in hypothyroid animals, we decided to extend our observations to the pyramidal cells of the hippocampus in order to gain further insight into the effects of hypothyroidism upon the other neuronal links of the hippocampal trisynaptic circuitry, inasmuch as CA1 neurons are known to be particularly vulnerable to aggressive agents. Groups of 6 male and 6 female rats aged 30 and 180 days were analysed separately after being treated as follows: (1) hypothyroid from day 0 until day 30 (30-day-old hypothyroid group); (2) respective 30-day-old control; (3) hypothyroid from day 0 until day 180 (180-day-old hypothyroid group); (4)hypothyroid until day 30 and thenceforth maintained euthyroid (recovery group); (5) hypothyroid since day 30 (adult hypothyroid group); and (6) respective 180-day-old control. The volume of the pyramidal cell layer of the CA1 and CA3 regions and the numerical density of the respective neurons were evaluated, thereby allowing us to estimate the total number of pyramidal cells in each hippocampal region. The areal density and the mean nuclear volume of CA1 and CA3 pyramidal cells were also estimated. In the CA3 region, we found that hypothyroidism, whatever its duration and time of onset, induces a reduction in the volume of the pyramidal cell layer and a parallel increase in the numerical density of its neurons, without interfering with the total number of pyramidal cells. Conversely, in the CA1 region, thyroid hormone deficiency started either neonatally or during maturity was found to lead to a decrease in the total number of pyramidal cells. Reductions ranging between 14.2 and 22.5% were found in 30 and 180-day-old hypothyroid groups. The reestablishment of a euthyroid state did not ameliorate the referred neuronal loss. The present results support the view that hypothyroidism induces small alterations in the structural organization of the hippocampal CA3 region, contrary to what happens in CA1 in which neuronal death occurs. Furthermore, the data presented herein demonstrate that the total number of CA1 pyramidal cells displays sexual dimorphism that is not affected by thyroid hormone manipulations. These findings allow us to admit that hypothyroidism induces anatomical changes in the hippocampus which are in line with those reported for the dentate gyrus, thereby providing additional morphological basis for the cognitive alterations described in this condition. o 1992 Wileg-Liss, Inc. Key words: thyroid deficiency, hippocampus, pyramidal cells, neuronal death, sexual dimorphism
In humans, the exposure to low levels of thyroid hormones during late fetal or early postnatal periods commonly leads to developmental abnormalities and mental deficiency (Burgi, '86; Fisher, '89). The relationship between cognitive deficits and thyroid hormone deficiency became apparent after the report Of Kerley ('36) about childhood myxedema. Subsequently, additional work reinO
1992 WILEY-LISS. INC.
forced this view by further demonstrating that the cognitive and sensory alterations that are hallmarks of congenital hypothyroidism are either irreversible or only partially Accepted ~ ~29, 1992, ~ i l Address reprint requests to M.D. Madeira, Department ofAnatomy, Porto Medical School, Alam. Prof. Herngni Monteiro, 4200 Porto, Portugal.
M.D. MADEIRA E T AL. reversible, even when appropriate replacement therapy is promptly started (Hulse, '84; Biirgi, '86; Fisher, '89). The search for clues to the nature of this dramatic situation justifies the intensive research that, during the past decades, has been conducted in several regions of the central nervous system (CNS) under hypothyroid conditions. From the wealth of work produced, it became evident that the presence of normal levels of these hormones is indispensable for the harmonious development and subsequent maintenance of the structural organization of the CNS. Most of the morphological work produced in this domain was performed in the cerebellum (Nicholson and Altman, '72a,b; Legrand, '82-'83; Madeira et al., '88a) and in the neocortex (Eayrs and Taylor, '51; Eayrs, '55; RuizMarcos et al., '79, '83). In contrast, studies designed to evaluate the effects of thyroid hormone deficiency upon the structure of the hippocampal formation are scarce, which is hard to understand, since the hippocampal formation plays a role of paramount importance in a wide range of cognitive functions that are known to be deeply affected in hypothyroidism, namely learning processes and memory tasks (Strehler, '89; Alkon et al., '911, especially those including spatial information (McNaughton et al., '89; Alkon et al., '91). Experiments that used simple learning tests (Eayrs, '61; Eayrs and Levine, '63; Davenport and Dorcey, '72) succeeded in demonstratingthat in animal models, hypothyroidism exerts deleterious effects upon learning capacities; i.e., it induces behavioral alterations similar to those observed in humans. This is likely because most of the morphological alterations observed in the CNS of hypothyroid animals parallel those found in human studies performed in patients with thyroid hormone deficiency (Gilroy and Meyer, '75). The hippocampal formation is particularly suitable for morphometric studies because it is a relatively simple cortical structure (West et al., '78; Stanfield and Cowan, '79; Amaral and Witter, '89). Still, the existing quantitative studies regarding the effects of hypothyroidism upon its morphological organization are almost exclusively restricted to the granule cells of the dentate gyrus (Rami et al., '86b; Madeira et al., '88b, '91b), which are known to have a peculiar neurogenesis (Bayer and Altman, '74; Bayer, '80; Altman and Bayer, '90a). We have unequivocally demonstrated that hypothyroidism, starting either in the neonatal period or in adulthood, leads to a reduction in the total number of these neurons, which is not restored by the normalization of thyroid hormone levels (Madeira et al., '88b, '91b). We were able to conclude that the mechanisms that underlie those changes were multiple: in addition to the widely accepted interference with granule cell migration (Lauder, '79; Legrand, '82-33; Rami et al., '86b), thyroid hormone deficiency also alters granule cell proliferation and induces iieuronal degeneration (Nicholson and Altman, '72a; Legrand, '82-'83; Madeira et al., '91b). Bearing in mind that granule cells are just the first link of the hippocampal circuitry (Amaral and Witter, '89), we thought it would be worthwhile to extend our observations, using the same experimental paradigm, to the remaining components of the aforementioned circuitry. Thus, we centered our studies on pyramidal cells of hippocampal regions CA1 and CA3 that display a neurogenic pattern that is markedly different from that of granule cells as it occurs predominantly during the prenatal period (Bayer and Altman, '74; Bayer, '80; Altman and Bayer, '90b). Neurogenesis is o f paramount importance for the understanding of
hypothyroid-induced alterations because in neurons generated prenatally, the alterations mainly derive from changes in the process of differentiation and thus are confined to cell neurites, becoming apparent on the components of the neuropil (Ruiz-Marcos et al., '79, '83; Madeira et al., '90, ' 9 1 ~ ) Conversely, . in neuronal populations that undergo a significant postnatal formation, besides those alterations (Legrand, '82-'83; Rami et al., '86a), a reduction in the number of neurons is likely t o occur, since both neuronal proliferation and migration are also affected by the hormone deficiency and thereby interfere with the process of cell acquisition (Rami et al., '86b; Madeira et al., '88a,b, '91b). The estimators chosen to be evaluated were the total number of CA1 and CA3 pyramidal cells, their respective numerical density, and the volume of the pyramidal cell layer, both in young (30-day-old) and adult (180-day-old) animals submitted to different experimental conditions of thyroid hormone deficiency. These determinations were undertaken separately for CA1 and CA3 regions since their respective pyramidal cells exhibit a selective vulnerability to various aggressions (Nadler et al., '78; Griffiths et al., '84; Bertram et al., '90; Represa et al., '91; SchmidtKastner and Freund, '91 ). Moreover, as hippocampal neurons display receptors for gonadal hormones (Stumpf and Sar, '78; Loy et al., '88) and hypothyroidism alters the entire hormonal milieu (Dumm et al., '851, we attempted to analyse whether the total number of pyramidal neurons displays sexual dimorphism either under normal conditions or following thyroid hormone manipulations. With this quantitative investigation, an extension of the study undertaken in the dentate gyrus under hypothyroid conditions, we expect to get a better insight into the effects of hypothyroidism in the hippocampal formation by answering the following questions: (1)does hypothyroidism that starts during the neonatal or adult periods interfere with the number of pyramidal cells?, (2) does long-term hypothyroidism induce progressive cell death?, (3) is there any morphological improvement once euthyroidism is reestablished?, (4)is there any regional vulnerability to hypothyroidism among CA1 and CA3 pyramidal cells?, and (5) are there sex differences among CA1 and CA3 pyramidal cells in normal and hypothyroid conditions?
MATERIALS AND METHODS Animals and treatments Some of the animals herein employed were used in a previous work (Madeira et al.,'91b). Data were collected from 30 and 180-day-old SpragueDawley rats from the colony of the Gulbenkian Institute of Science (Oeiras, Portugal). Males and females were separately analysed. On the day of birth (day PO) the litters were equalled to 8 (4male and 4 female). Separate groups of 6 male and 6 female rats, formed by pooling animals from at least 5 different litters, were treated as follows: Thirty-day-old hypothyroid groups. Hypothyroidism was induced from day PO until day P30 by a subcutaneous daily injection of propylthiouracil (PTU): 0.05 mlO.2% PTU on days PO-P10, 0.1 mlO.2% PTU on days Pll-P20, and 0.1 mlO.4% PTU on days P21-P30 (Nicholson and Altman, '72a; Madeira et al., '91b). One-hundred eighty-day-old hypothyroid groups. Rats were treated as in the previous group from day PO untjl day P30. At day P30, a thyroidectomy was surgically performed
HYPOTHYROIDISM AND HIPPOCAMPAL PYRAMIDAL NEURONS with the aid of a dissecting microscope after intraperitoneal anaesthesia (2 mlikg body weight) with a solution of 2.5% sodium pentobarbital in physiological saline (Madeira et al., ’91b). Recovery groups. Hypothyroidism was restricted to the first postnatal month. The newborn rats were rendered hypothyroid from day PO until day P30 by administering PTU as previously described. Thence, they were allowed to survive without further treatment. Adult hypothyroid groups. Rats were maintained in standard conditions as the controls until day P30 and then hypothyroidism was induced by surgical thyroidectomy. Thirty and 180-day-old control groups. Rats were not submitted to any kind of treatment during both experimental periods. For the sake of consistency, each experimental group should have its respective control group subcutaneously injected with physiological saline and/or sham-operated, as animal manipulations might increase the blood levels of stress hormones known to exert deleterious effects upon neurons (Sapolsky, ’87). However, the glucocorticoidinduced neuronal loss in the hippocampal formation only occurs following cumulative hormonal exposure over the lifespan (Sapolsky, ’87). Thus, we decided not to include those controls because enlarging the number of groups to be studied would render very difficult to get an overall view of the effects of hypothyroidism. In addition, our own experiments (Madeira et al., ’91b), as well as data from other authors (Matsumoto et al., ’911,clearly show that this kind of animal manipulation does not alter the number of the hippocampal neurons. As males and females were separately studied, 12 groups of 6 animals each were analysed. The animals were housed at a light-dark cycle of 12/12 hours and after weaning they had free access to standard laboratory diet and water. Body weights were recorded every third day during the first month of experiment, and thereafter every fifteen days; the last determination was performed on the day of sacrifice.
Hormonal determinations Serum thyroxine (T4) levels were determined by radioimmunoassay by using 50 pl blood samples and commercial kits supplied by Pharmacia. Though the assays of triiodothyronine, reverse triiodothyronine, and thyroid-stimulating hormone would be desirable for a full characterization of hypothyroidism, they were not performed because specific immunoreactants were not commercially available. Yet, T4 concentration is an accurate estimator of hypothyroidism and thus the obtention of a complete profile of the thyroidal hormone status was not considered to be an absolute requisite.
General procedures At the end of each experimental period (30 or 180 days), animals were anaesthetized with ethyl ether and, after heparinization, blood samples (1 ml) were collected from the heart for hormonal assays. The rats were then transcardially perfused with a solution of 1%glutaraldehyde and 1% paraformaldehyde in 0.12 M phosphate buffer at pH 7.2. The thyroidectomized rats (180-day-old hypothyroid and adult hypothyroid groups) were carefully searched for the presence of thyroid gland remnants and when any evidence of thyroid tissue was present, the animals were discarded.
503
In the remaining groups, the thyroid glands and the adjacent portion of the trachea were isolated and weighed. The brains were removed from the skull and weighed. The cerebral hemispheres, after being separated by a longitudinal section in the midsagittal plane, were immersed for 2 hours in fresh fixative solution. The left hippocampal formations were isolated by dissection from the adjacent structures and weighed.
Volumetric estimations After removal of the frontal and occipital poles of the right hemispheres, the blocks of tissue containing the hippocampal formations were dehydrated in an ascending series of ethanoliethanol-ether and embedded in celloidin (Fig. 1A). After the embedding procedure, the blocks were serially sectioned in the horizontal plane at a nominal thickness of 60 pm with a sliding microtome. The sections from each hippocampal formation were mounted in glass slides and stained with cresyl violet (Fig. 1B). The actual section thickness, evaluated by differential focusing with a x 100 oil immersion objective lens, was found to be 59.3 ? 6.4 pm. The volume of the pyramidal cell layer of the CA1 and CA3 regions of the hippocampus was estimated from sections covering the entire dorsoventral extent of the hippocampal formation (Fig. 1B). Because the morphology of the hippocampal components changes markedly from section to section at both septal and temporal poles, the first 25 septal and the last 10 temporal sections were studied. At the midseptotemporal level, where the profile of the hippocampal components changes very little, sections studied were 240 pm apart; i.e., only every fifth section was selected for volumetric determinations. The boundaries of the pyramidal cell layer were delineated according to the criteria described in detail by West et al. (’78). Although the terminology used to designate the hippocampal regions followed in this paper is the one described by Lorente de N6 (’341, in the pyramidal cell layer only two regions were considered: CA1 and CA3 (Fig. 1B).The distal border of the first one was considered at the point where the densely packed pyramidal cell layer gives rise to the wider cell layer of the subiculum (Figs. lB, 2A,B). The discrimination between CA1 and CA3 was made taking as reference the CA2 region (Fig. 2A,C) which is easy to define by the abrupt transition in cell size and cell layer organization (Lorente de NO, ’34).For volumetric estimation purposes, this small transitional zone was included in CA3 because the transition between CA2 and CA3 is hardly detectable in conventionally stained sections. In addition, the neurons located in the pyramidal cell layer within the hilus of the dentate gyrus, which can be clearly differentiated from the hilar cells (Lorente de N6, ’34; Amaral, ’781,were also considered as belonging to the CA3 region. From all the selected sections, the boundaries of the pyramidal cell layer of the CA1 and CA3 regions were drawn using a camera lucida attachment at a magnification of x63 and their areas were measured with a MopVideoplan. Based on these measurements, the volume of the pyramidal cell layer of both hippocampal regions was calculated according to the Cavalieri’s principle using the “arithmetic mean” approximation given by the equation: n-l
V = S F , i=x1
A,
+ Ai+l
di,
M.D. MADEIRA ET AL.
504 in which SF, is the tissue shrinkage factor, A, is the surface area of the respective pyramidal cell layer in the iJh section, n is the number of sections measured, and d, is the distance between areas A, and A,+1(Uylings et al., '86). The SF, was estimated separately for male and female rats at the septal and temporal poles of the dissected left hippocampal formations. After removing the midseptotemporal part of the hippocampal formations, a section was immediately obtained in the horizontal plane and by using an Oxford vibratome from the surface of each pole facing the site of transection; the cross-sectional area of these sections was then measured. The remaining portions of both poles were embedded in celloidin, and the area of the first section obtained from both blocks, immediately adjacent to the vibratome section, was also measured. The referred areal measurements were made with the aid of a Mop-Videoplan after photographing the respective sections on a stereoscopic microscope. The SF, was then calculated from the areas of the two immediately adjacent sections according to the formula:
in which c is a shape constant, A, is the vibratome section area, A, is the celloidin section area, and V, and V, are the volumes of the sections obtained with the vibratome and after embedding in celloidin, respectively (Uylings et al., '86).
Quantifications of neuronal parameters
sections selected at random from different blocks and cutting 3 cross-sections of each at the same thickness. Three measurements were made in each transverse section and the mean thickness determined was 2.02 2 0.20 pm. All neuronal estimations were performed in the subregion c of CA3 (Lorente de NO, '34) and in the proximal part of CA1 (Fig. 2A). However, because the methodology applied was identical for control and experimental groups, there is no reason to assume that this might have interfered with the end results of a comparative study such as this. Mean nuclear uolume. The size of the pyramidal cell nuclei was estimated using the nucleator (Gundersen, '88; Gundersen et al., '88; Madeira et al., '91b). Photographs of the same area of the pyramidal cell layer of the CA1 and CA3 regions were obtained from sets of semithin sections of the hippocampal formation, and then analysed at a final magnification of ~ 1 , 8 6 0 (Fig. 2D). The nuclei to be measured were selected by applying the disector (Sterio, '84; Madeira et al., '88a,b, '91b). The estimations were carried out in three-dimensionally isotropic directions using a systematic set of points (Cruz-Orive and Hunziker, '86; Gundersen et al., '88) and a bidirectional ruler equidistant in length cubed (Gundersen and Jensen, '85). Taking into account the ruler construction and the magnification of the photographs (Braendgaard and Gundersen, '861, the mean volume of the nuclei),W was estimated according to the equation:
4T -
VN -- -3e ; ,
For neuronal quantifications, the left hippocampal formations were used. They were sectioned in the horizontal plane using a tissue chopper and the blocks processed, at room temperature, as though being prepared for electron microscopic observation (Palay and Chan-Palay, '74). After postfixation in a solution of 2% osmium tetroxide in 0.12 M phosphate buffer, the blocks were rinsed in 25% ethanol and then dehydrated through a graded series of ethanols. Block-staining took place at the 70% stage with 1%uranyl acetate during 1 hour. After passage through propylene oxide the material was embedded in Epon. Because the neuronal density and size vary throughout the septotemporal axis of the hippocampal formation (Gaarskjaer, '781, and because there is evidence that under some experimental conditions the vulnerability of hippocampal neurons changes along the septotemporal axis (Ashton et al., '89), a systematic random sampling procedure (Gundersen and Jensen, '87) was applied to select the sections used for the quantification of neuronal parameters. Therefore, 6 blocks were selected from each left hippocampal formation, and froin each block, 2 groups of 3 consecutive 2 pm semithin sections were obtained, mounted in glass slides, and stained with toluidine blue. The actual section thickness was determined after reembedding 10 semithin
in which is the mean of all cubed distances measured (Gundersen, '88). Areal density. The number of cells per unit surface area of the pyramidal cell layer-areal density (NA)-was calculated by counting, at a final magnification of ~ 3 2 0 the , number of existing neurons within the boundaries of a previously determined area of the pyramidal cell layer in the CA1 and CA3 regions. For this purpose, the first and the last sections of each group of 3 consecutive semithin sections were used per animal. Numerical density. The number of cells per unit volume of the pyramidal cell layer-numerical density (Nv)was estimated by applying the disector (Sterio, '84; Madeira et al., '88a,b, '91b). From all groups of semithin sections obtained, photographs of the same area of the pyramidal cell layer of the CA1 and CA3 regions were taken from two 0 consecutive sections at a final magnification of ~ 3 2 (Fig. 2D). A total of 24 disectors was made per animal. Total number of cells. The total number of cells of the hippocampal CA1 and CA3 regions was calculated by multiplying the volume of the pyramidal cell layer of each region, after correction for SF,, by the numerical density of the respective neurons. For technical reasons, it was not possible to estimate these parameters in the same hippocampal formation, which is known to be of great impor-
Fig. 1. A: Lateral view of the right hippocampal formation from a 180-day-old male control rat after removal of the neocortex. The lines (a,b,c,d)indicate the location from where the sections present in B were obtained. x 5 . 3 . B: Light micrographs of horizontal celloidin sections taken from representative levels of the right hippocampal formation whose location is indicated by the lines traced in A. The borders
between CA1 and CA2-3 regions are indicated by arrows, whereas the borders between CA1 and the subiculum are indicated by arrowheads The CA1 and CA3 regions of the pyramidal cell layer from those sections can be seen, at a lower magnification, in the correspondent schematic camera lucida drawings. CA1 and CA3, regions of the hippocampus; DG, dentate gyrus; S, subiculum. x 18.4.
Figure 1
A
Figure 2
HYPOTHYROIDISM AND HIPPOCAMPAL PYRAMIDAL NEURONS tance to get unbiased results. Still, for comparative purposes, the results obtained can be considered reliable because the methodology employed was the same for all groups studied. With the histological processing used, the discrimination between pyramidal and nonpyramidal cells is impossible to be achieved and, therefore, the results obtained are estimates of the total number of neurons in the pyramidal cell layer of each hippocampal region. Since pyramidal cells are the predominant cell type in the pyramidal cell layer and the nonpyramidal neurons account just for a very small proportion of the cells in this layer (Lorente de N6, '34; Schlander and Frotscher, '86; Danos et al., '91), it is reasonable to assume that the results obtained are representative mainly of the former cell type.
Statistical analysis Normal quantile plots were used as a tool for assessing the normal distribution and the variance homogeneity of the data (Moore and McCabe, '89). Due to the normality and variance homogeneity of the sampled populations, a single two-way analysis of variance (ANOVA) was carried out on data from control and experimental groups to discern main effects. Treatment and sex were used as independent variables and the animals as replicates. The remainder mean square was used as the error term. Post hoc linear polynomial contrasts (Moore and McCabe, '89) were applied to test whether group means differed significantly from each other; painvise comparisons were performed. A Multistage-Bonferroni test (Moore and McCabe, '89) was used to control for unacceptable levels of type I error, thus obtaining a stronger protection for false rejections of the null hypothesis. Results for which the null hypothesis would be rejected according to the uncorrected procedure applied are regarded as borderline. Throughout this study, values are expressed as mean 5 S.D.
RESULTS Body weights
507
TABLE 1. Results of Two-way ANOVA on the Body, Brain, Hippocampal Formation, and Thyroid Gland Weights of the 30-Day-OldAnimals Treatment (df 1,20)' F value Body weight Brain weicht Hippocamid formationweight Thyroid gland weight
Sex (df 1,201
P value
Interaction (df 1,201
F value P value F value P value
132.72 5 1 39
